Satellites flying at 360km (the target altitude for starlink V3) deorbit very quickly without regular burns. Dead starlink satellites are guaranteed to come down within 5 years.
Just coverage is already provided by the 600-something direct to cell satellites already in orbit yes, but you need more if you want it to be useful beyond loading text-only posts or sending SMS
One difference is you usually only get one shot to manipulate a human before they get suspicious. If the LLM's context resets you can try all over as if your first failed attempt didn't even happen.
It was more that IT considered the college-hosted git instance "experimental" still, in multiple project courses we were required to use college hosted version control as part of the class, but those who wanted to use git were on their own to figure it out.
When I went to college (early 2010s) professors were still encouraging students to use SVN, so I probably have a fair number of peers who didn't learn git until they got out into the real world
If you're big enough you can get direct access to Nvidia engineers, and they are usually transparent when they find out the bug was in their software and send you a patched version to try to resolve the issue
When you run CUDA at scale dealing with nvidia driver and library bugs takes up a disgustingly large percentage of engineer time, I don't know a lot of people who would be looking forward to rely on more nvidia libraries.
Interesting, I just tried it and it got server-side the first try, although it wants to insert a hyphen. I did download the larger dictionary from their site.
I'm somewhat lucky in that I didn't have to be the one to force that issue, another friend in my gaming group already made the decision that he would switch to Linux and no longer play any games that did not work on Linux/Proton. So it was pretty easy for me to just switch last year.
Not sure how much data they retained but it's possible they also have the _input_ tokens, and in the "early" days (how far everything moves in just a couple years...) more of those input tokens would have been non-AI-generated
I've been burned by this outside games as well. Computation heavy process, in production it writes the inputs into the output as well for reproducibility, but some of them used unordered maps so it would slightly differ when you loaded it back up. Long term solution was to stop using unordered maps in general, but short term we had to make sure the inputs got sorted before being inserted so they would have the same insertion order every time...
The Nvidia GB300 DGX Station, which isn't even going to hit 1TB total memory, is expected to launch at almost $100k. Bit of a pipe dream with memory prices where they're at.
If you live in most places in the US other than the urban heart of a few very large cities you have to take a huge hit to your ability to get places in a reasonable time frame without a car. I have hope some more cities other than NYC are improving the situation, but as it is the closest I got to using public transit for a commute was when I was going to one of our other offices in a different downtown area I would drive my car to the park n ride to take the train the rest of the way. The train saves time and sanity because traffic downtown is a nightmare, but that drive takes 5 minutes, and it would add 20+ minutes if I had to walk to the closest bus stop so I could take the bus up to the train station.
The same could be said for a lot of drugs that have strong side effects, just one difference here is the main "side effects" are something some people consider fun. Though you could still do double-blind for comparing the effectiveness of different psychedelics, e.g comparing LSD to Psilocybin.
Only a problem if you're trying to use AI to forgo creating a user interface for untrusted users (probably the worst idea that's seeing widespread use right now)